Gas Consumption in France

Big Data Analysis of Gas Prices using Apache Spark

Project Overview

This project analyzes 3 years of gas price data (2022-2024) from over 14,000 gas stations across France using Apache Spark. We build a complete data pipeline: from data collection to price forecasting.

Objectives

  • Load and process gas price data with PySpark
  • Visualize gas price evolution over time
  • Build a machine learning model to forecast next-day prices
  • Create geographic visualizations of price indexes by department

Dataset

File Description File’s size
Prix2022S1.csv Prices Jan-Jun 2022 ~178MB
Prix2022S2.csv Prices Jul-Dec 2022 ~156MB
Prix2023.csv Prices 2023 ~322MB
Prix2024.csv Prices 2024 ~310MB
Stations2024.csv Station locations ~2.6K
Services2024.csv Station services ~980K

The total price records is 14,214,837

Source: French Government Open Data via GitHub

Gas Types Analyzed

  • Gazole (Diesel) - Most common
  • SP98 (Premium Unleaded 98)
  • E10 (Unleaded with 10% ethanol)
  • E85 (85% ethanol - flex gas)

Results

1. Weekly Price Evolution (2022-2024)

Gas Price Evolution
Gas Price Evolution

Key Observations:

  • 2022 Energy Crisis: Sharp price spike around week 20-25 (Probably Ukraine war impact)
  • Peak prices: ~2.20 €/L for SP98, ~2.15 €/L for Gazole
  • E85: Consistently cheapest (~0.80-1.15 €/L)
  • 2023-2024: Gradual stabilization around 1.70-1.90 €/L
  • Price hierarchy: SP98 > E10 > Gazole > E85

2. Geographic Analysis - Price Index by Department

  • Interactive maps showing which departments are cheaper (green) or more expensive (red) than the national average. Dowmload the following files to visualise it
Gas Type Map
Gazole View Map
SP98 View Map
E10 View Map
E85 View Map
  • Static map

    Gazole
    SP98

    E10
    E85

Price Index Interpretation:

  • = 1.0 -> Station price equals national average
  • > 1.0 -> More expensive than average (red)
  • < 1.0 -> Cheaper than average (green)
  • In general, Gas is more expensive in Paris and its surroundings
  • E85 and E10 is highly expensive in Corse

3. Price Forecasting Model

We built models to predict next-day gas prices using lag features (past prices as inputs).

This prediction uses the Gazole gas .

Model Comparison

Model R2 RMSE MAE
Linear Regression 0.9569 0.0163 0.0121
Random Forest 0.9241 0.0216 0.0152

Dispersion Plot (Actual vs Predicted)

Model Dispersion
Model Dispersion

Interpretation: - Points close to the red diagonal = good predictions - Linear Regression performs slightly better (R2 = 0.9569)

4. Feature Importance

Feature Importance
Feature Importance

Most Important Features: 1. price_lag_1 (35%) - Yesterday’s price is the best predictor 2. price_lag_2 (25%) - Price 2 days ago 3. price_rolling_7d (19%) - 7-day moving average 4. price_lag_3 (14%) - Price 3 days ago 5. price_lag_7 (4%) - Same day last week

Least Important from random forest: day_of_week, month, day_of_month (< 2% combined)

Feature Importance LR
Feature Importance LR

Conclusion: Gas prices are highly autocorrelated - recent prices are the best predictors of future prices. Calendar features have minimal impact.

Technical Stack

Tool Purpose
PySpark Big data processing (Billion of records)
Spark SQL Data aggregation and queries
Spark ML Machine learning pipelines
Matplotlib/Seaborn Static visualizations
Folium Interactive geographic maps
Pandas Data manipulation for plotting

Project Structure

├── data/
│   ├── Prix2022S1.csv
│   ├── Prix2022S2.csv
│   ├── Prix2023.csv
│   ├── Prix2024.csv
│   ├── Stations2024.csv
│   └── Services2024.csv
├──figures/
|   |---gas_prices_evolution.png         # Price evolution chart
|   ├── model_dispersion_plot.png        # Model evaluation
|   ├── feature_importance.png           # Feature importance
|   ├── france_gas_prices_map*.html      # Interactive maps
├── gas_consumption_france.ipynb    # Main notebook
├── config.yaml                      # Configuration file
└── README.md

How to Run

Prerequisites

pip install numpy pandas pyyaml pyspark matplotlib seaborn geopandas folium

A java version installed

Execution

  1. Download data files from GitHub
  2. Place CSV files in ./data/ directory
  3. Run the Jupyter notebook: gas_consumption_france.ipynb

Key Formulas

Price Index

Price Index = 100 × (Station Price - National Average) / National Average + 1

Week Index

Week Index = floor((Current Date - First Date) / 7) + 1

👤 Author

Glorie METSA WOWO
January 2026

📄 License

This project uses open data from the French government.